On the real effects of private equity investment: evidence from new ...

Working Paper SeriesNo 1078 / august 2009On the Real Effectsof Private EquityInvestmentEvidence fromNew Business Creationby Alexander Popovand Peter Roosenboom

WORKING PAPER SERIESNO 1078 / AUGUST 2009ON THE REAL EFFECTSOF PRIVATE EQUITYINVESTMENTEVIDENCE FROM NEWBUSINESS CREATION 1by Alexander Popov 2and Peter Roosenboom 3In 2009 all ECBpublicationsfeature a motiftaken from the€200 banknote.This paper can be downloaded without charge fromhttp://www.ecb.europa.eu or from the Social Science Research Networkelectronic library at http://ssrn.com/abstract_id=1436894.1 We thank Marco Da Rin, Philipp Hartmann, Ulrich Hege, Florian Heider, Josh Lerner, Simone Manganelli, Marina Martynova, Eric Nowak,Enrico Perotti, Jose Luis Peydro, Richard Rosen, Per Stromberg, Krishnamurthy Subramanian, and Gregory Udell, as well as seminarparticipants at the ECB, the 18 th IT&FA conference, the 2008 EFA meeting, the RICAFE2 third conference, the 11 th Symposiumon Finance, Banking and Insurance, the 2009 MFA meeting, and the Federal Reserve Bank of Chicago for valuable comments.A special thanks is extended to Luc Laeven for providing us with the Amadeus and D&B data. We also thank Lieven Baertand Kim Bonnema for outstanding research assistance. The opinions expressed herein are those of the authorsand do not necessarily reflect those of the European Central Bank or the Eurosystem.2 European Central Bank, Financial Research Division, Kaiserstrasse 29, D-60311 Frankfurt am Main, Germany;tel. +4969 1344 8428, email: Alexander.Popov@ecb.europa.eu3 Department of Finance, Rotterdam School of Management (RSM), Erasmus University Rotterdam,Burgemeester Oudlaan 50, NL 3062PA Rotterdam, The Netherlands;tel: +3110 408 1255; e-mail: proosenboom@rsm.nl

© European Central Bank, 2009AddressKaiserstrasse 2960311 Frankfurt am Main, GermanyPostal addressPostfach 16 03 1960066 Frankfurt am Main, GermanyTelephone+49 69 1344 0Websitehttp://www.ecb.europa.euFax+49 69 1344 6000All rights reserved.Any reproduction, publication andreprint in the form of a differentpublication, whether printed orproduced electronically, in whole or inpart, is permitted only with the explicitwritten authorisation of the ECB or theauthor(s).The views expressed in this paper do notnecessarily reflect those of the EuropeanCentral Bank.The statement of purpose for the ECBWorking Paper Series is available fromthe ECB website, http://www.ecb.europa.eu/pub/scientific/wps/date/html/index.en.htmlISSN 1725-2806 (online)

CONTENTSAbstract 4Non-technical summary 51 Introduction 72 Data 102.1 EVCA yearbooks 102.2 Amadeus database 122.3 US industry-level entry data 142.4 Sample construction 153 Empirical methodology 174 Results 184.1 Private equity and entry: main results 184.2 Private equity and entry: contemporaneousvs. long-term effect 204.3 Alternative measures of propensity to entry 214.4 Private equity and access to finance 234.5 Endogeneity and selection 254.6 Robustness 285 Conclusion 30References 33Tables 37Appendix 46European Central Bank Working Paper Series 49ECBWorking Paper Series No 1078August 20093

AbstractUsing a comprehensive database of European firms, we study how private equityaffects the rate of firm entry. We find that private equity investment benefits newbusiness incorporation, especially in industries with naturally higher entry rates andR&D intensity. A two standard deviation increase in private equity investment explainsas much as 5.5% of the variation in entry between high-entry and low-entry industries.We address endogeneity by exploiting data on laws that regulate private equityinvestments by pension funds. Our results hold when we correct for barriers to entry,general access to credit, protection of intellectual property, and labor regulations.Keywords: private equity, venture capital, firm entryJEL Classification: G24, L26, M134ECBWorking Paper Series No 1078August 2009

Non-technical summaryRecent years have spurred interest in the role of private equity investment in thefinancing of small new firms. While banks are often reluctant to finance such firmsbecause of high uncertainty, information asymmetry, and agency costs, private equityinvestors are specialized to overcome these problems through the use of stagedfinancing, private contracting and active monitoring. These unique features make themmore likely to finance early stage and technology companies than banks. This paperinvestigates the previously unexplored effect that private equity investments have onnew business creation in Europe. This question is highly relevant to policy makersgiven that they often perceive venture capital as an important contributor to the risingleadership of US firms in high technology industries. Hoping to rival this success, theEuropean Union stimulates venture capital investment in an attempt to make Europe ahotbed for entrepreneurship.Our study contributes to the literature in two main ways. First, although there is alarge body of empirical literature looking at the effect of finance on firm entry, the vastmajority of it has studied the impact of banks on business formation. While the effectof banking sector development and restructuring on small business formation iscertainly important, the effect of private equity investment, especially start-up finance,is probably equally important and largely understudied. Secondly, while academicinterest in venture capital has been growing steadily, there is still a remarkably limitedresearch on the effects of private equity and venture capital on the real economy.Several recent studies have argued that venture capital can be credited with stimulatinginnovation, spurring entrepreneurial spawning, and enhancing productivity growth.However, most of this interest has focused on US evidence. This paper thus providesthe first cross-country cross-industry study of the effect of private equity investment onnew business creation in Europe, using data from Amadeus on entry and data from theEuropean Venture Capital Association on firm creation. The data also enables us todistinguish between stages and investor types and to include the effects of later stageprivate equity, like buyout finance.We first study if private equity investment, aggregated as well as by stagedistribution, affects the extent of incorporation, using data on 1998-1999 and on 2006-2007. We aim at identification by following a difference-in-differences methodologybased on cross-country cross-industry interaction effects. This approach allows us tobypass the omitted variables problem that has plagued traditional research bycontrolling for unobservable characteristics of the industries of interest as well as thebusiness environment in the respective country. We find that the rate of incorporationin naturally high-entry and R&D-intensive industries is significantly higher incountries with a larger volume of private equity investment relative to GDP, and this isparticularly true for smaller firms.We conduct extensive robustness tests to correct for endogeneity and for the classicindustry- and country-level determinants of new business creation suggested by theliterature. While previous studies have show that regulations the firm needs to meet inorder to be registered as a limited liability company explain a large portion of thisvariation, we show that private equity finance is also a very important determinant offirm entry when entry barriers are accounted for. Further accounting for access tofinance in general and for other regulatory and legal characteristics of the businessECBWorking Paper Series No 1078August 20095

environment leaves the main results unchanged. Finally, we account for the possibilitythat these results are driven by reverse causality. We use the variation in prudentialregulation of the investment behavior of pension funds as an instrument for the supplyof PE funds. Our results prove robust to this IV procedure.This paper serves to shed light on the contribution of private equity investment tofirm creation in Europe. We show that VC’s contribution to creative destruction in theEuropean context is very tangible, and has an important role alongside regulatory, tax,and labor market reforms in promoting a dynamic EU economy. In that context, avariety of policy initiatives, undertaken in recent years and aimed at promoting anactive venture capital industry, are expected to stimulate further Europe’s younginnovative firms.6ECBWorking Paper Series No 1078August 2009

1 IntroductionIt is generally accepted that access to credit is an important determinant of …rm entry andgrowth (Rajan and Zingales (1998)). However, banks are often reluctant to …nance smallnew …rms because of high uncertainty, information asymmetry, and agency costs (Beck et al.(2005)). Private equity investors are specialized to overcome these problems through the useof staged …nancing, private contracting, and active monitoring (Hellmann (1998); Gompersand Lerner (1999, 2001a); Kaplan and Stromberg (2001)) and are therefore more likely to…nance early stage and technology companies than banks.In this paper, we investigatethe previously unexplored e¤ect that these private equity investments have on new businesscreation in Europe. This question is highly relevant to policy makers given that they oftenperceive venture capital as an important contributor to the rising leadership of US …rmsin high technology industries (Gompers and Lerner (2001b)). Hoping to rival this success,the European Union stimulates venture capital investment in an attempt to make Europea hotbed for entrepreneurship (Aernoudt (1999); Gilson (2003)). This paper puts the ideathat venture capital fosters new business creation to the test.Our study makes two key contributions to the literature. First, although there is a largebody of empirical literature looking at the e¤ect of …nance on …rm entry, the vast majorityof it has studied the impact of developments in the banking sector on business formation,with mixed results. Petersen and Rajan (1995) show that …nancial liberalization hurts smallyoung companies because creditors in competitive credit markets …nd it more di¢ cult tointernalize the bene…ts of assisting those …rms. However, Black and Strahan (2002) …ndevidence that the rate of incorporation increases as a result of banking deregulation, and ina related paper Cetorelli and Strahan (2006) use a measure of bank deregulation to show howaverage …rm size in an industry decreases with bank competition, arguably due to increasedentry rates. Regarding general credit market development, Aghion et al. (2007) …nd thatdeeper and more developed banking sectors are associated with higher entry of small …rmsin sectors which are more dependent on external …nance. While the e¤ect of banking sectorECBWorking Paper Series No 1078August 20097

development and restructuring on small business formation is certainly important, the e¤ectof private equity investment, especially start-up …nance, is probably equally important andlargely understudied. This paper aims to …ll this gap.The second contribution is that we add to a remarkably limited research on the e¤ectsof private equity on the real economy. Kortum and Lerner (2000) show that venture capitalinvestment in the United States is associated with more innovation as measured by patentcounts and patent citations. Gompers et al. (2005) examine the propensity of publicly traded…rms to create new venture backed …rms. They …nd that younger public …rms located inmain hubs of venture capital activity are the most likely to create new ventures. Tang andChyi (2008) …nd that venture capital investment enhances productivity growth. Regardingnew business formation, we are only aware of one study that examines the impact of venturecapital at the regional level in the U.S. Mollica and Zingales (2007) report that …rm entryand innovation increase in U.S. regions that attract more venture capital. Our paper addsto this literature by providing the …rst comprehensive cross-country study examining thee¤ect of private equity on …rm entry in Europe. Moreover, our European data enables usto distinguish between stages and investor types and does not focus on the e¤ect of venturecapital alone but also includes the e¤ects of later stage private equity, like buyout …nance.There are two main mechanisms suggested by the literature via which private equity ingeneral and venture capital in particular should lead to higher rates of business incorporation.First, nascent entrepreneurs may recognize the need for capital in the future andonly establish …rms when they have reasonably high expectations of obtaining such funding.This implies that not just start-up …nance, but later …nancing stages, like expansion …nance,should matter too for …rm entry. Second, …rms may be engaged in "entrepreneurial spawning",that is, the propensity of former employees of publicly traded …rms to start their owncompanies. There are several competing explanations for that. Christensen (1997) has arguedthat large, established …rms are incapable of adopting radical new technologies becauseit would disrupt their established way of organizing business. It is also possible that these…rms cannot evaluate new disruptive technologies as they fall outside of their line of business8ECBWorking Paper Series No 1078August 2009

(Stein (2002)). Firms sometimes choose to not adopt new technologies although they can,because this would lead to a decline in the productivity of their existing businesses (Schoar(2002)). In all of those cases, venture capital would be the obvious …nancing tool for thenew …rms born out of these technologies. Finally, Gompers et al. (2005) …nd evidence thatentrepreneurial spawning is related to the fact that employees of established …rms are trainedand conditioned to be entrepreneurs by being exposed to the entrepreneurial process and byworking in a network of entrepreneurs and venture capitalists. It needs to be emphasizedthat in this paper we only examine empirically the e¤ect of private equity on new businesscreation, rather than studying the exact channels via which this e¤ect works.The literature has distinguished entry into an industry from …rm creation.The …rstaccounts for the migration of …rms across industries, while the second emphasizes pureentrepreneurship (de novo …rms). We focus on the second approach and de…ne entry as theincorporation of a previously nonexistent …rm in the respective industry and country. Ourdata comes from Amadeus, a comprehensive database of corporations across a number ofdeveloped and transition countries in Europe, in combination with country-level data onprivate equity and venture capital investment in Europe from the European Venture CapitalAssociation (EVCA) yearbooks.We …rst study if the volume of private equity investment, aggregated as well as by stagedistribution, a¤ects the extent of incorporation, using data on 1998-1999. We aim at identi-…cation by following the di¤erence-in-di¤erences methodology …rst introduced by Rajan andZingales (1998) and focus on cross-country cross-industry interaction e¤ects. In essence, westudy whether the fraction of new incorporation is higher in an industry with higher "natural"entry rates when there is more private equity ‡owing into the country. This approachallows us to bypass the omitted variables problem that has plagued traditional research bycontrolling for unobservable characteristics of the industries of interest as well as the businessenvironment in the respective country. We …nd that the rate of incorporation in naturallyhigh-entry industries is signi…cantly higher in countries with a larger volume of private equityinvestment relative to GDP, and this is particularly true for smaller …rms. The same appliesECBWorking Paper Series No 1078August 20099

to industries which are more R&D intensive.We also look at staging and investor type, distinguishing between VC and buyouts, aswell as between independent, captive, semi-captive, and public funds.We …nd that thee¤ect of venture capital and start-up …nance is comparable to the e¤ect of total privateequity …nance, but that it was stronger in 2006-2007 than it was in 1998-1999. We also …ndthat while cross-country di¤erences in private equity investment by independent funds isboth economically and statistically signi…cant, the e¤ect of captive and public funds on newbusiness entry is at best non-existent. While Klapper et al. (2006) show that regulationsthe …rm needs to meet in order to be registered as a limited liability company explain alarge portion of this variation, we show that private equity …nance is still a very importantdeterminant of …rm entry when entry barriers are accounted for.In a …ne-tuning of the basic approach, we account for the possibility that these results aredriven by reverse causality, that PE is a proxy for other types of …nancial development, andthat it is a proxy for other characteristics of the business environment. We use the variationin prudential regulation of the investment behavior of pension funds as an instrument for thesupply of PE funds. Our results are robust to this IV procedure, as well as to accounting foraccess to …nance in general and for other regulatory and legal characteristics of the businessenvironment.The paper proceeds as follows. In Section 2 we summarize the data. Section 3 describesthe empirical methodology. Section 4 presents the empirical results. Section 5 concludeswith the main …ndings of the paper.2 Data2.1 EVCA yearbooksThis paper uses data from two main sources: on new business formation from the …rmlevelAmadeus database, and on private equity investment from the EVCA yearbooks. TheEVCA yearbooks compile annual data on private equity funds raised, funds allotted to10 ECBWorking Paper Series No 1078August 2009

venture capital, and the actual allocation of private equity investment. We use the data forthe 1998-1999 period, and in robustness tests for the 2006-2007 period.Three caveats are in place. First, while the EVCA yearbooks try to be exhaustive interms of the European countries they cover, in some cases they discontinue their reporting(e.g. Iceland after 2001). In others - notably, the new EU members from Central Europe(Czech Republic, Hungary, Poland and Slovakia) - EVCA only started reporting PE activityin 1998.Finally, for the Baltic and South-East European EU members EVCA has onlyrun pilot projects limited in duration (Bulgaria in 2003 and 2004, Latvia in 2001, Romania2000-2004, and Slovenia in 2003), or reported private equity investment jointly for severalcountries (Croatia and Slovenia, as well as Estonia, Latvia and Lithuania in 2005). Understandably,in cases when there were too few years included, or when it was judged impossibleto disaggregate reliably the information on private equity, the data was not used. Apart fromcurrent EU members, the EVCA yearbooks also include information on Iceland, Norway, andSwitzerland.The second caveat deals with the reporting of investment by US private equity houses. Ifa deal has been backed by both a US and a European private equity house, the deal is splitinto two parts. The part of the investment coming from the European private equity …rm isallocated to the respective European country, and the part of the investment coming fromthe US private equity …rm is allocated to the US. However, if the US PE …rm has no o¢ cein Europe, then its investment is not included in the EVCA …gures. In addition, concerningUS private equity houses investing in Europe, only investments made by those having o¢ cesin Europe are taken into consideration. This would imply that if a US PE …rm, which hasno o¢ ce in Europe, invests in a European company, the investment would not be includedin the EVCA …gures. While the vast majority of US private equity houses operate throughtheir European o¢ ces, it is still the case that the EVCA data is by construction incomplete.Finally, while EVCA o¤ers disaggregated data on the allocation of actual private equityinvestment across industries, it uses its own classi…cation of 17 groups of industries, and itonly reports the composition of investment by country of management rather than countryECBWorking Paper Series No 1078August 200911

of destination.We …rst recalculated investment by industry and country of destinationassuming the same pattern applies across industries as across aggregate volumes 1 . Next, weused the special translation key provided by the EVCA to translate the EVCA industriesinto 2-digit NACE rev. 2 and then into 2-digit NACE rev. 1.1 industries in order to be ableto match them to the Amadeus database. In some cases the NACE two-digit industries fallunder two or more di¤erent EVCA industries at the same time, and this double countingwas resolved by assigning the 2-digit NACE industry to only one of the two or more EVCAindustries interchangeably or dropping those altogether. The …nal results reported are robustto using di¤erent variants of this industry translation key.Table 1 summarizes the information on total actual private equity investment normalizedby GDP, both for the 1998-1999 and the 2006-2007 period.It gives a clear idea of thevolatility of private equity investment both in aggregate and in relative terms. For example,total investment has decreased by a magnitude of 2 in the Czech Republic (from 0.042 to0.023) and by a magnitude of 6.5 in Greece (from 0.027 to 0.004), but it has almost tripledin the UK (from 0.465 to 1.215), quadrupled in Sweden (from 0.2 to 0.82) and increased bya magnitude of 12 in Denmark (from 0.029 to 0.353). It is important to note that our resultsare robust to these country-level developments.2.2 Amadeus databaseThe …rm-level data come from the Amadeus database.Amadeus is a commercial pan-European database provided by Bureau van Dijk, containing …nancial information on over10 million public and private companies in 38 European countries. It combines data fromover 30 specialist regional information providers (IPs). The data is created by collectingstandardised annual accounts (for up to 10 years), consolidated and unconsolidated, for approximately9 million companies throughout Europe, including Central, Eastern, and South-1 EVCA reports aggregate PE investment by country of management and country of destination and PEinvestment in each industry by country of management only. We recalculate industry PE investment bycountry of destination assuming that the ratio between total and per-industry investment is the same forcountry of management and for country of destination. For the 2006-2007 wave, however, the disaggregateddata come by country of destination.12 ECBWorking Paper Series No 1078August 2009

Eastern Europe. The database contains detailed …rm-level accounting data for a numberof …nancial ratios, activities, and ownership. While initially received from over 50 di¤erentvendors across Europe, the data is then transformed into a single format enabling comparisonacross countries. The focus of the Amadeus database is on …nancial information, like…rm pro…t, revenue, assets, debt, and value added. In addition to that, Amadeus provides…rm-level information on year of incorporation and employment. We use the former to calculatethe age of the …rm, hence the share of "new" …rms in each industry-country-year. Thevariable we create is referred to as Entry ij , and it denotes the share of …rms less than 2-yearsold in country i in industry j, calculated separately for 1998-1999 in the main regressions,and for 2006-2007 in the robustness tests. We only count the …rms that are at least 1 fullyear of age to reduce measurement error.Finally, Amadeus uses the 3-digit NACE industry classi…cation standard, which we aggregateat the 2-digit level in order to have a su¢ cient number of …rms in each industry foreach country. Columns 2 and 4 of Table 2 summarize our main Amadeus data, aggregatedat the country level, for the two time periods of interest.One main concern is that the years 1998-1999 chosen for the main empirical exercisesmay not be "steady state" in terms of new business incorporation as they were too closeto the peak of the dot-com bubble. Similar concerns apply to the period 2006-2007 whichcoincided with the peak of another business cycle. For that reason, we calculated averageentry rates over the 1995-2007 period, using data on entry rates from Eurostat, and thencompared the long-term averages to data for 1998-1999 and 2006-2007, again from Eurostat.Columns 3 and 5 of Table 2 presents the deviation of the latter from the long-term averagesin percentage terms, for 1998-1999 and 2006-2007, respectively. The average deviation forthe 1998-1999 sample is 5.9%, and only in three of the countries in the sample is it higherthan 7%. This gives us con…dence that the years 1998-1999 are pretty much "steady state"in terms of new business creation. The same applies to the second period chosen. 22 The little overall variation in entry rates is signi…ed by the fact that even the years 2000 and 2001, whichcoincided with the peak of the internet bubble, show an average deviation of only 8.4% from the samplelong-term average.ECBWorking Paper Series No 1078August 200913

2.3 US industry-level entry dataAs a benchmark for new business incorporation into an industry, we use data on …rm entryin the respective NACE rev. 1.1 2-digit industry in the US, from the Dun and Bradstreetdatabase of over 7 million corporations over the period 1998-1999.The methodology ofusing US industry characteristics as a benchmark in cross-country cross-industry studieswas …rst introduced by Rajan and Zingales (1998), who argued that the composition ofUS industries in terms of external …nance usage can be viewed as the industries’"natural"or "technological" composition, because US …nancial markets are relatively friction-free,compared with other …nancial markets around the world, including most industrial countries.The methodology has been used, among others, by Beck et al.(2008) to calculate the"natural" share of small …rms in an industry, by Claessens and Laeven (2003) to calculatethe industry’s "natural" usage of intangibles, and most recently by Klapper et al. (2006) tocalculate the "natural" rate of business incorporation, which we use in this paper.It needs to be pointed out that proxying the "natural" rate of industry entry by entryin the respective industry in the US is somewhat arbitrary. It relies on the assumption thatbureaucratic barriers to entry are lower there than in any country in Europe, and that thecost of …nancial intermediation and start-up …nancing is su¢ ciently lower in the US than inEurope as a whole. This technique does not argue that industries in the US have achievedthe …rst best in terms of entry, but that the …nancial and regulatory environment is relativelymore conducive to entry than the countries in our sample. For example, while entry costs inthe US are around 0.5% of per capita GNP, in our sample of European countries they areon average around 20%; and while venture capital investment in the US in 1998-1999 wasabout 0.2% of GDP, it stood at 0.1% of GDP in the UK, 0.05% of GDP in France and 0.02%of GDP in Germany, with these three countries receiving the lion share of the private equityinvestment in Europe. Certainly, the entry costs in the US are non-zero, and it is impossibleto know if the optimal amount of venture capital investment is not much larger than 0.2%of GDP, but what matters is that these characteristics of the US business environment aresuperior than in the countries in our sample.For robustness purposes, however, similar14 ECBWorking Paper Series No 1078August 2009

to Claessens and Laeven (2003), we replace the US industry characteristics with industrycharacteristics from other regions in some speci…cations.In Table 3, we compare the industry entry rates in the US to an aggregated measureof European entry from the Amadeus-EVCA dataset. In more than half of the industries,entry in the US is at least marginally higher than in Europe as a whole.2.4 Sample constructionWe employ the same sample selection procedure to the Amadeus dataset as Klapper et al.(2006) to see whether the e¤ect of private equity investment survives the inclusion of barriersto entry in a sample where these barriers were already shown to be signi…cant. We initiallyuse the 2001 edition of Amadeus and limit the sample to the years 1998 and 1999. The year2000 is not used due to some incompleteness of the information in each Amadeus editionconcerning the previous year. The years prior to 1998 are not used due to the well-knownsurvivorship problems of the Amadeus database: when a …rm ceases existing, Amadeus keepsa record of it for 4 years, and then takes it out of the database. Consequently, while eachyearly addition of Amadeus contains data for more than 4 years back, the sample of …rmsone will …nd reported for the time-period 4 years and more prior to the year of issuing, willnot include many …rms who existed in that year, but exited the market after that. Using thedata indiscriminately will therefore induce survivorship bias and misrepresent the volume ofentry, and so we focus our attention to the years 2 and 3 prior to the year the database wasissued. In later robustness tests, in which we use the 2008 edition of Amadeus, we similarlyfocus our attention on the sample of …rms in 2006-2007.Firms in the …nal dataset we use are also required to have basic accounting informationon the variables we are interested in (year of incorporation, employment, assets, etc.). Thisapproach excludes phantom …rms created for tax purposes. We drop …rms that report onlyconsolidated statements in order to avoid double-counting …rms and subsidiaries abroad.We exclude industries where the activities are country-speci…c, namely agriculture, forestry,…shing, and mining.We also exclude utilities and post and telecommunications, whichECBWorking Paper Series No 1078August 200915

tend to be heavily regulated and/or state-owned, and the …nancial services sector because,arguably, …nancial …rms are subject to speci…c regulations which do not apply for other …rms(for example, initial capital requirements). Finally, we exclude the public sector, education,the social sector, private households, and activities that cannot be classi…ed. We are leftwith 37 NACE industries.At the country level, we exclude the East European and South-East European countriesdue to insu¢ cient availability of data on investment by private equity, although thesecountries are covered by Amadeus. The same applies to Luxembourg. Likewise, we excludeSwitzerland, which has comprehensive coverage on private equity investment, but itsAmadeus coverage is compromised by the fact that small …rms are not required to …le. Finally,we use Eurostat to con…rm to what extent Amadeus is representative of the …rm sizedistribution in the respective country. We then exclude Iceland, Ireland, and Portugal, forwhich the ratio of employment in …rms with more than 250 employees in Amadeus to employmentin …rms with more than 250 employees in Eurostat is less than 0.5, and/or forwhich the di¤erence between the share of small …rms (10-50 employees) in Amadeus and inEurostat is more than 0.25. The sample thus reached represents the best match of Amadeusand EVCA data that is possible to construct while avoiding limited coverage, insu¢ cientobservations and country-speci…c industry scope and legal requirements problems. The …nalsample consists of 2,788,680 …rms for 1998-1999 in 16 countries: Austria, Belgium, Czech Republic,Denmark, Finland, France, Germany, Greece, Hungary, Italy, Netherlands, Norway,Poland, Spain, Sweden, and UK. Firm-level information is then aggregated at the industrylevel and matched with the EVCA data to create a dataset consisting of 571 2-digit NACErev. 1.1 industry-country data points (37 industries in 16 countries, with 21 data pointsmissing). Due to our sample selection method, we do not use 3 of the 17 EVCA industries(other electronics, …nancial services, and agriculture).Table 4 gives the conversion key from the 2-digit NACE rev 1.1 industries into the EVCAindustries used in robustness regressions. Note that out of the 37 NACE rev. 1.1 industries,12 fall exclusively under 1 EVCA class, for 10 more than 70% of the 3-digit subclasses16 ECBWorking Paper Series No 1078August 2009

fall under 1 EVCA class, and for 15 industries there is a clear "winner" (majority of 3-digitclasses), but still the possibility of a measurement error remains. In these empirical exercises,we use di¤erent ways of matching the industries in the third group to EVCA classes, and wealso perform the analysis after excluding them altogether. The reported results are robustto this method.Our sample selection procedures leaves us with the same sample as in Klapper et al.(2006) in terms of business entry. This allows us to test the e¤ect of private equity onbusiness entry accounting for the e¤ect of entry barriers in a sample in which the e¤ect ofentry barriers has already been documented.3 Empirical methodologyWe use a cross-industry cross-country regression as our main empirical model:Entry ij = 0 + 1 (EV CA i(k) Entry US ) + 2 (X i Z j ) + 3 D i + 4 D j + " ij (1)where Entry ij denotes the share of …rms less than 2-years old in country i in industry j;EV CA (i)k denotes private equity investment in country i, or in EVCA industry class k incountry i, normalized by the country’s GDP and measured as total volume or disaggregatedby stage; Entry US denotes the "natural" industry entry rate, as measured using US data;X i is a vector of country characteristics; Z j is a vector of industry characteristics 3 ; D i is amatrix of country dummies; D j is a matrix of industry dummies; and " ij is the idiosyncraticerror.This speci…cation has become very popular in the …nance and growth literature because italleviates the small sample problem of cross-country regressions, and it allows for eliminatingthe e¤ect of unobservables characteristics of the business environment by including countrydummies. Industry indicator variables are also included to account for unobservable industry-3 We use j to denote NACE Rev. 1.1 industry class and k to denote EVCA industry class as they are notidentical, but they can be matched.ECBWorking Paper Series No 1078August 200917

speci…c e¤ects. Critically, we are interested in the magnitude and statistical signi…cance ofthe estimate of 1 . This interaction term measures the e¤ect of private equity investment onnew business incorporation accounting for the industry’s "technological" entry rate. Privateequity investment is expected to make it easier for …rms to enter the market, and that e¤ectis expected to be larger in industries in which there are high "natural" entry rates. Hence,we expect the sign of 1 to be positive. Note that we are not able to identify the direct e¤ectof PE investment and of the industry’s "natural" entry rate, as those are fully captured bythe two sets of …xed e¤ects.It has become regular practice in the literature to account for potential convergence e¤ects(that is, the fact the larger sectors may have naturally lower entry rates) by controlling forthe industry share (see, for example, Cetorelli and Strahan (2006) and Beck et al. (2008)).We proxy this by the fraction of industry sales out of total sales in the country as reportedin the Amadeus database.Finally, in the main empirical exercise, we use average data for 1998-1999. Later, forrobustness purposes, we perform the same estimation on data from 2006-2007.4 Results4.1 Private equity and entry: main resultsIn Table 5, column (i) we present the basic OLS regression when private equity investment (asreported by EVCA) is aggregated by country of destination. The coe¢ cient on the interactionterm is signi…cantly positive, implying that relative entry into industries with naturallyhigher entry is disproportionately higher in countries with large levels of private equityinvestment, normalized by GDP. What the coe¢ cient means numerically is the following:let’s take a high entry-industry (at the 75th percentile of entry) and a low-entry industry(at the 25th percentile), and let’s take two countries which score high (75th percentile) andlow on private equity (25th percentile). The two countries in mind are Hungary (low) andDenmark (high), with the di¤erence between the two being 0.859.Then, the estimated18 ECBWorking Paper Series No 1078August 2009

coe¢ cient implies that the di¤erence in entry rates between the high-entry and the lowentryindustry are 0.44 percentage points higher in Denmark than in Hungary. Thus, allelse equal, an entrepreneur envisioning starting a …rm in a high-entry industry is to a largerdegree better o¤ by operating in Denmark rather than in Hungary than an entrepreneurwho wants to enter a low-entry industry. It is also useful to think of it in terms of the actualvariations in entry rates between the 75th percentile and a 25th percentile industry which is8%. Hence, private equity accounts for 5.5% of that di¤erence.This simple empirical test shows that …rst, private equity investment has a real e¤ectthrough …rm creation. 4Second, the magnitude of increasing private equity investment bytwo standard deviations in our 16-country sample is about half of the e¤ect of loweringentry barriers by two standard deviations in the 24-country sample used by Klapper et al.(2006). 5 In later tests, we perform horse race regressions to compare the e¤ect of privateequity against the e¤ect of barriers to entry, property rights protection, tax burden, andother characteristics of the business environment, and in all cases private equity remains asigni…cant predictor of the variation in cross-country cross-industry entry rates.In the rest of the columns of Table 5, we test the basic result using di¤erent methodologiesand sample speci…cations. In column (ii), we account for left and right censoring by replacingthe OLS speci…cation with a Tobit one. The rationale behind this is that entry rates areleft-truncated at 0 and right-truncated at 1. The coe¢ cient of the interaction term does notchange, however, it becomes statistically signi…cant at the 1% level. Next, we exclude thesub-sample of transition countries (column (iii)). The 1990s were a vibrant period in theeconomic history of Central Europe in terms of privatization, and we would like to eliminatethe possibility that a large number of new private …rms are actually old state-owned …rmswhich have been counted only after they became private.Our results are robust to theexclusion of these countries, but the magnitude of the coe¢ cients and the resulting numericale¤ect decreases slightly. We also look at the e¤ect of private equity on incorporation of small4 However, before addressing the endogeneity probelm, we will be using the word "e¤ect" with caution.5 Klapper et al. (2006) …nd that entry barriers explain about 10% of the entry di¤erence between highandlow-entry industries.ECBWorking Paper Series No 1078August 200919

…rms only (less than 10 employees, column (iv)). The reason for that is that our e¤ect so farcould be upward biased by the fact that some of the "new" …rms in our sample may in factbe the result of M&A. We …nd that the di¤erence in small …rm entry rates between a highanda low-entry industry (75th percentile vs. 25th percentile) is 4.6%, and so a coe¢ cient of0.263 implies that di¤erences in private equity investment explains about 9.5% of the meandi¤erence. We can thus conclude that private equity has a real e¤ect on …rm creation, andthis e¤ect is strongest for small new …rms.Finally, we replace our variable measuring entry rates with a variable measuring exit rates.Entry and exit are considered complementary in creative destruction theories (Schumpeter(1942); Aghion and Howitt (1992; 1998); Geroski (1995)). We use data from Eurostat on theshare of …rms who dropped from the sample at time t but were in it at time t1 to proxy forthe exit element of the creative destruction hypothesis. The implicit hypothesis being testedis that entry rates are also a¤ected by unobservable variables like entrepreneurial culture andrisk attitudes, and so observing an e¤ect of private equity investment on entry rates mayoverstate the true e¤ect of private equity by having the true e¤ect contaminated by demandconsiderations. Exit rates are much more likely to be determined by the supply of fundsrather than the demand for them; one most likely channel for that e¤ect is that ine¢ cientincumbent …rms will leave the population of …rms faster under pressure from more e¢ cientPE-backed newcomers. The last column of Table 5 con…rms that private equity investmentworks well along the lines of the theory of creative destruction, with its e¤ect on exits beingeconomically and statistically at least as strong as the measured e¤ect on entry.4.2 Private equity and entry: contemporaneous vs. long-terme¤ectThere are a number of robustness checks we need to perform on our measure of private equity.First, we have only regressed so far new business entry in 1998-1999 on the contemporaneousmeasure of private equity investment. Alternatively, we could do so using an average measure20 ECBWorking Paper Series No 1078August 2009

of private equity investment over several years back. Such an approach would make sure thatthe estimated e¤ects are not biased by a temporary idiosyncratic shock to private equityinvestment, especially given the degree of ‡uctuation of investment highlighted in Table 1.It is also conceivable to hypothesize that new entrants pay more attention to the long-termstate of the private equity industry rather than to current investment because long-terminvestment patterns usually convey more information about the future availability of privateequity …nance. For example, the business registration fees and the initial investment areusually …nanced with own funds or with …nancial help from the "three F" (family, friends, andfools), and hence the incentive to start a company which will later be …nanced with venturecapital or expansion money, or outright bought by a private equity investor, is provided bythe long-term presence of private equity investors rather than the current state of the market.Indeed, our evidence suggests that the long-term presence of private equity also associatedwith higher relative entry than contemporaneous one (Table 6). In fact, for both total new…rms and new …rms with less than 10 employees, the e¤ect of longer-term private equity iseven higher, suggesting that it goes beyond the direct a¤ect of actually …nancing the seedand start-up stages of the …rm’s life. This is along the lines of the theories outlined in theintroduction, with new …rms not only being born out of old relationships between venturecapital and entrepreneurship, but also from the expectation of getting …nance in later stagesvia expansion …nancing.4.3 Alternative measures of propensity to entryNext, we account for the possibility that our measure of entry is not the best proxy ofpropensity to entry. Prior literature (e.g. Gerotski (1995)) has suggested that exit rates area good proxy for propensity to entry because higher …rm creation is necessarily associatedwith higher …rm destruction.In the …rst two columns of Table 7, we use the Dun andBradsteet data to calculate the average share of exiting …rms for 1998-1999, per 2-digitindustry. US industry exit rates turn out to be as good proxies for propensity to entry asUS entry rates, in the sense that the e¤ect of private equity investment on entry is relativelyECBWorking Paper Series No 1078August 200921

higher for industries with higher exit rates, repeating the results of the previous estimation.However, the result only holds when we perform the estimation for the new …rms with lessthan 10 employees only (column (ii)).We next make use of the fact that new …rms are generally small because entry by larger…rms may re‡ect M&A activity. As much as it applies to the rest of the world, it should alsoapply to our benchmark US case. We therefore replace the original proxy for "natural" entrywith a measure of the average industry entry rates over 1998-1999 for small and mediumenterprises (SMEs), or …rms with less than 250 workers. When we apply that methodology,the estimate of 1 remains positive, implying relatively higher entry due to private equityinvestment in industries that have higher entry by SMEs in the US. The estimates aresigni…cant at the 5% level. This result holds for entry by new …rms in di¤erent size classes.Finally, we use Claessens and Laeven’s (2003) insight that it is not necessary to use USdata to calculate "natural" industry characteristics, as long as we use benchmark data froma business environment that greatly outperforms the countries in the dataset in terms ofbusiness opportunities, …nance, and regulations. They show that using UK and Hong Kongdata to calculate "natural" intangibles usage yields identical results as when using the USbenchmark. We test that prediction by using UK data from Amadeus rather than US oneto calculate industry entry rates. This methodology also accounts for the fact that what weassume are "natural" or "technological" entry rates could be driven by peculiarities of the USindustry structure. Using a UK benchmark makes sure that we are pinning entry in Europein general against an industrial structure which is more oriented towards manufacturing thanthe US, and against a legal system which is more creditor-friendly than the US (La Portaet al. (1998)). We exclude from the regressions the industry data points from the UK. Theregression results con…rm that entry in the UK and entry in the US are similar benchmarksfor natural propensity to entry. Our estimates for 1 are signi…cant at least at the 1%, butonly when we look at total new …rms.22 ECBWorking Paper Series No 1078August 2009

4.4 Private equity and access to …nanceAfter accounting for the possibility that our main proxy for "natural" propensity to entry isimperfect, we next account for the possibility that private equity investment is a mere proxyfor other types of …nance or for return to investment. For instance, countries with highervolumes of private equity investment will tend to have better developed banking sectors, soour measure of private equity will be capturing the e¤ects of the credit market on entry viabusiness loans. Also, countries with dynamic PE and VC industries will tend to have higherinvestor protection, and so again our volume measure will be contaminated by the e¤ect ofthe expected return on investment on entry. Most importantly, as Perotti and Volpin (2007)argue, the volume of …nance may matter less than access to …nance per se. Hence, our PEmeasure might be picking up the e¤ect on new business entry of easier access to all kinds of…nance, including consumer loans and mortgages. In all of those cases, our estimates wouldbe biased.Therefore in Table 8 we proceed to measure the e¤ect of private equity investment onentry alongside the e¤ect of …nance in general. In column (i), we show the estimates ofa regression which includes a measure of private credit by commercial banks, normalizedby GDP. This measure is widely accepted as a good proxy for a range of …nancial issues,like access to business loans, depth of the …nancial sector a¤ecting the ability of …nancialplayers to gain access to investment opportunities, etc. (Rajan and Zingales (1998); Becket al. (2008)). The e¤ect of private credit on new business entry is both economically andstatistically signi…cant, as expected.Importantly, the e¤ect of private equity investmentremains signi…cant, although in the case of all …rms its magnitude decreases somewhat.However, in the case of …rms with less than 10 employees, the e¤ect of private equity in newbusiness incorporation remains as strong as before (column (v)).Next, we account for the fact that private credit is also a volume measure and thus animperfect proxy for access to …nance. Therefore, we employ a formal proxy for access to…nancial services (columns (ii) and (vi)) taken from the World Bank’s "Finance for All?Policies and Pitfalls in Expanding Access" which is a composite indicator measuring theECBWorking Paper Series No 1078August 200923

percentage of the adult population with access to an account with a …nancial intermediary.While this index captures more than access to business loans, it is a better measure thanthe volume of private credit of how easy it is to access …nancial services in general. Thecorrelation between the two measures is 0.71, implying that they are highly but not perfectlycorrelated, and so the formal index could indeed be capturing more access issues than privatecredit. Again, while it is correlated with new business incorporation, including it in theregression doesn’t eliminate the e¤ect of private equity on …rm entry. While in the case oftotal new …rms the estimate of 1 again loses part of its economic and statistical signi…cance,it remains the case that the volume of private equity …nance impacts business creationindependent of general access to …nancial services.We also look at investors’protection. Rajan and Zingales (2003) argue that the absenceof regulation protecting investors could be a very e¢ cient barrier to new …rm creation. Theright measure of …nancial development, the argument goes, would capture not only the easewith which any entrepreneur or company with a sound project can obtain …nance, but also thecon…dence with which investors anticipate an adequate return. The previous two measureswe used would then be a poor proxy for this investor con…dence, and we next proceed toincorporate in our regression a direct measure of the degree to which individual investmentsare protected by the legal system in the country. 6The indicator we employ is a compositeof the quality of three indices: transparency of transactions, liability for self-dealing, andshareholders’ability to sue o¢ cers and directors for misconduct. As expected, this index hasa very signi…cant e¤ect on entry when interacted with our measure of "natural" entry rates,pointing to the fact that investors indeed take into account the degree of legal protectiona¤ecting the expected return to individual investments in start-up companies.Tellingly,the e¤ect of private equity investment on business entry survives this extension of the basicmodel (columns (iii) and (vii)).Finally, we do a horse race in which we include all country-level measures used in Table8 so far interacted with our measure of natural industry entry (columns (iv) and (viii)).6 Perotti and Volpin (2007) also use this measure as a proxy for access to …nance.24 ECBWorking Paper Series No 1078August 2009

We …nd that private credit doesn’t enter signi…cantly anymore, and access to …nance isinsigni…cant in the case of entry of all new …rms, but in general, our results con…rm thatthe depth of …nancial system, access to …nance and investor protection all matter for entry.Importantly, the private equity interaction continues to enter positively and statisticallysigni…cantly, albeit with a somewhat decreased order of magnitude.4.5 Endogeneity and selectionThe empirical methodology chosen is traditionally prone to endogeneity problems. A measuredpositive coe¢ cient on the composite term of interest does not automatically implycausality; it could be that private equity is endogenous to …rm entry rates, or it couldbe that a set of omitted variables is jointly driving both the propensity to enter and thepropensity to invest in start-up companies, in interaction with our industry characteristicsof choice. The traditional solution in this line of research is to use an instrumental variables(IV) procedure to account for this potential endogeneity. It has been generally agreedthat the country’s legal origin is a strong predictor of the degree of legal regulation and thequality of the …nancial system nowadays (La Porta et al. (1998)). However, in the caseof private equity and entry there are two problems which may reduce legal origin to anine¢ cient instrument. For one, it could be that the link between the exogenous componentof the legal system and private equity investment is relatively weak, but more importantly,it is conceivable that legal systems a¤ect entry via channels other than private equity investment,like barriers to entry, for instance. Hence, we rather look at the types of lawsthat were in place in the 1990s, regulating the ability of pension funds managers to engagein private equity investment. The method is akin to Kortum and Lerner (2000), who use a1979 clari…cation of the Employee Retirement Income Security Act (ERISA) "Prudent man"rule by the Department of Labor allowing pension funds to engage in PE investments as aninstrument for VC investment. 7The idea is that the increased supply of funds reduces the7 Gompers and Lerner (1999) show that VC investment increases more than …ve-fold following the ERISA"Prudent man" rule clari…cation by the US Department of Labor.ECBWorking Paper Series No 1078August 200925

cost of …nancing to venture capitalists, and hence serves as a supply shifter, while it has noe¤ect on the availability of ideas that need start-up …nance. 8We create a dummy equal to 1 if pension funds were allowed to invest in private equityin the respective country prior to the sampling period, and to zero otherwise. We also createa variable which is constructed by interacting the pension funds dummy with the averagesize of pension funds in the respective country as share of GDP over the 1995-1999 period 9 .Being correlated with the predetermined components of private equity and venture capitalinvestment and exogenous to current business opportunities makes these variables reasonableinstruments, even in a small-sample context with linear estimation. 10In an (unreported)exercise we …nd that a …rst-stage regression including the two instruments one at a time, aswell as together, explains between up to 65% of the cross-country variation in private equity,with an F-value of up to 12.5. In addition, countries that enacted such a measure have seenthere PE funds raised increase by 95.5% more on average over the period 1991-1999 thancountries which did not. We report the estimates from the second stage in columns (i)-(ii)and (vi)-(vii) of Table 9. We …nd that the estimate of 1 is still positive and still highlysigni…cant, and does not vary dramatically in magnitude from the estimates of the OLSregressions.However, a common ‡aw of this method is that countries with large industries withhigh natural entry may have both higher entrepreneurial culture (resulting in more entryregardless of private equity investment) and higher levels of private equity investment (dueto higher demand for all types of …nance). One method to control for that possibility is toexclude the industries that are in the right tail of the industry size distribution. We restrictour sample to the industries that are in the bottom tertile, bottom two tertiles, or outside thetop 10% of their country’s industries in terms of size. When we do that, in all cases we get8 Unfortunately, given the aggregated data we are provided with, it is impossible to construct a moredirect proxy of the cost of …nance.9 The data for pension funds size in the 1990s comes from Eurostat.10 It is possible that these laws were enacted following pressure from a growing PE industry, or a lack ofentrepreneurial activity (which would make them correlated with PE and hence null the exclusion restriction).It is a fact, however, that the formal motive expressed during the legislative process has universally beendiversi…cation of risk (EVCA yearbooks, 1991-1999).26 ECBWorking Paper Series No 1078August 2009

estimates which are still positive and signi…cant, similar to magnitude to the ones estimatedpreviously, and they do not vary much regardless of the de…nition of "small industries" thatwe use. The coe¢ cient after the exclusion of the top 10% is reported in columns (iii) and(viii) (for entry by all …rms and by …rms with less than 10 employees, respectively).Another selection concern is that in countries whose culture is adverse to entrepreneurialactivity (and hence, there is naturally less entry), there are also stricter rules guiding privateequity investment, as the local legislatures would be adverse to business activity in general.We account for that possibility by distinguishing the general quality of laws in the respectivecountry. The above argument would imply that if it is a selection problem, private equityinvestment will have a larger e¤ect in countries where the quality of laws is generally lower.This is not the case; in fact as indicated in columns (iv) and (ix), the e¤ect is the opposite– more private equity leads to relatively higher entry in high-entry industries in countrieswith well-developed legal systems, implying no omitted variable bias.Finally, we have to account for the possibility that by focusing on legal entry, we may beoverlooking activity in the informal sector. To account for the possibility that our results arebiased by not considering a potentially large dynamics in the informal sector, we next useEnste and Schneider’s (2000) measure of the size of the informal economy to control for that.In columns (v) and (xi) of Table 9, we regress …rm entry on two triple interaction terms,in which the industry natural entry-country private equity investment has been interactedwith dummies for high and low degree of the informal economy, where low share of theinformal economy equals 1 if the country is in the bottom tertile of the informal economyshare, and to 0 otherwise, and high share of the informal economy equals 1 if the country isin the top tertile of the informal economy, and to 0 otherwise. The coe¢ cient on the tripleinteraction term including the low informal economy dummy is positive, implying that incountries where for tax or regulatory reasons there is little activity in the informal sector,private equity is indeed associated with higher entry in industries that naturally have higherentry rates.ECBWorking Paper Series No 1078August 200927

4.6 RobustnessWe conduct an extensive robustness analysis whose detailed exposition is left for a companionpaper available upon request for the sake of brevity. We …rst estimate the e¤ect of privateequity investment on entry, accounting for the standard industry determinants of entry thathave been suggested by the literature.In particular, it has been pointed out that entryrates are a¤ected negatively by …nancial dependence and capital intensity, and positivelyby technological opportunities and industry growth, among else. 11We therefore want tomake sure that the e¤ects we are measuring are not driven by other industry characteristicsfor which our measure of "natural entry" is a proxy. We …nd that some characteristics ofthe business environment matter for entry exactly as predicted: higher intellectual propertyprotection is associated with higher entry in industries that naturally invest more in R&D,and lower entry costs are associated with relatively higher entry in industries that are growingfaster. Importantly, in all cases higher private equity investment keeps its independent e¤ecton entry, and this e¤ect is still relatively stronger for naturally high-entry industries. Thise¤ect also survives the "horse race".Next, we look at a set of characteristics of the business environment, which the literaturehas identi…ed as predictors of entry: entry barriers, labor regulations, human capital, intellectualproperty rights protection, and taxes. To the degree that all of those tend to followa similar path over time, the e¤ect of private equity investment on entry could be contaminatedby developments along other dimensions of the business environment. We accountfor all those possibilities by including an interaction of the industry’s natural propensityto entry with another characteristic of the business environment, one at a time, and all ofthem simultaneously. We con…rm that barriers to entry have a detrimental e¤ect on entryrates. We …nd that labor regulations have no signi…cant e¤ect on entry, although Europe istraditionally regarded as a labor constrained environment. Better protection of intellectualproperty rights is associated with relatively more entry in high-entry industries, which alsotend to be the ones that are the most intangibles-intensive. Finally, cross-country variation11 See Geroski (1995) for a summary of the empirical evidence on that.28 ECBWorking Paper Series No 1078August 2009

in pro…t taxes does little to explain the cross-country cross-industry variation in entry ratesbetween high and low entry industries, and neither does a higher level of human capital oraccounting for the business cycle. Again, the e¤ect of private equity survives all robustnesschecks, including the horse race.As mentioned before, our data contains information on the distribution of private equityinvestment across 17 di¤erent industries as classi…ed by EVCA. This calls for a disaggregatedapproach in which we replace our measure of private equity investment at the country level(EV CA i ) with a measure of private equity investment at the country and industry level(EV CA ik ). The …rst di¢ culty is that the disaggregated data comes by country of management,unlike data on total private equity investment, which is reported both by countryof management and country of destination. We convert the industry data by country ofmanagement into country of destination by assuming that in each industry, the gap betweenprivate equity by country of management and by country of destination is the same as thegap between private equity by country of management and by country of destination for thewhole economy. This approach is obviously prone to measurement error and so the resultswe report should be taken with caution. The second di¢ culty deals with the issue of theNACE rev. 1.1 -> EVCA industry classi…cation translation key, highlighted in Section 2,for which we use the translation key developed and presented in Table 4. After repeatingthe estimation reported in Table 5, this time with disaggregated industry data, the e¤ect onnew business incorporation of the interaction of private equity with the industry’s "natural"entry rates are still economically and statistically signi…cant.Finally, we use the fact that the data from EVCA also allows us to account for thestaging of private equity investment (seed, start-up, expansion and replacement, and buyout),and for investment by type of investor (independent, captive, semi-captive, or publicfunds). The caveats of the previous paragraph apply again in the sense that the data onstaging and investor type comes by country of management only, and so our conversionof the data into country of destination relies on the assumption that the gap between, forexample, venture capital investment by country of management and by country of destinationECBWorking Paper Series No 1078August 200929

is the same as the gap between private equity by country of management and by countryof destination. However, in 2006 EVCA started reporting the stage distribution of privateequity …nance by country of destination too. Hence, in this last empirical exercise, we matchdata from the 2006-2007 EVCA yearbooks with data from Amadeus on entry for 2006-2007,using the 2008 CD. 12 We …nd that investor type matters: all the e¤ect of PE investmenton new business entry comes from investment by independent investors, while the e¤ect ofcaptive, semi-captive and public funds is at best non-existent. We also …nd that in 2006-2007,start-up …nance, VC and total private equity have an equally strong e¤ect on new businessincorporation. One way to read this is that nascent entrepreneurs anticipate future …nancingneeds and are more likely to decide to start their own business at a time when more venturecapital is available.A second explanation is o¤ered by entrepreneurial spawning whereventure-backed …rms are more likely to create other new ventures (Gompers et al., (2005)).However, this remains just a hypothesis until a more re…ned empirical approach looks intothe channels via which private equity investment a¤ects new business incorporation.5 ConclusionThis paper uses a cross-industry cross-country estimation technique in the spirit of Rajanand Zingales (1998) to identify the impact of private equity and venture capital investmentin general and start-up …nance in particular on entrepreneurship. We use two waves of theAmadeus database, which includes data on about three million …rms all across Europe: asample for the years 1998-1999 in the main empirical analysis, and a sample from 2006-2007for robustness purposes. The Amadeus represents one of the largest data sets on European…rms from these periods, and it is made particularly valuable by the fact that it includesextensive data on SMEs, private as well as publicly traded corporations, and is not limitedto manufacturing like many similar data sets. It also includes data on the age of the …rmand its year of incorporation, giving us the opportunity to proxy entry rates by the share of12 The same selection criteria apply as in Section 2.4.30 ECBWorking Paper Series No 1078August 2009

young …rms across the full range of NACE Rev 1.1 industries.We match that data to data from the European Venture Capital Association (EVCA) onprivate equity investment in general and start-up …nance in particular in Western Europe andsome Central European countries. We average the data from 1998-1999 to match with the…rst wave of Amadeus, and from 2006-2007 for the second wave. We examine 16 countries,37 NACE double-digit Rev. 1.1 industries, and 14 EVCA industry classes.We …nd that private equity investment has a bene…cial e¤ect on entry, which is relativelyhigher for industries which naturally have higher entry rates and are more R&D intensive.The e¤ect remains strong once we exclude investment allocated to buy-outs, suggesting thatearly stage …nance is important in this respect. Our results hold both in 1998-1999 and 2006-2007, when we account for industry size, and when we exclude the transition economies. Theresults stay unchanged after we address the endogeneity problem (does private equity induceentry or is it attracted to countries with a more dynamic industrial structure?) by usingan IV procedure in which variation in national prudential regulation guiding the investmentbehavior of Europe’s pension funds is used as an instrument for the supply of PE funds.We argue that by o¤ering a unique combination of ownership and incentives, private equityinvestment seems to lower the cost of start-up capital and result in higher industry dynamics.The results are generally robust to using di¤erent proxies for entry, contemporaneous orhistorical volumes of private equity investment, and to correcting for omitted variable bias.We also …nd that the e¤ect of private equity is higher in countries with better judicialsystems and in countries with smaller share of the informal economy. This result is onlylogical: to the extent that private equity targets industries that rely heavily on intangibleassets like intellectual property, licenses and patents, its e¤ect should be more pronouncedin countries where the returns to those are protected by the legal system. At the same time,we …nd that the relatively higher e¤ect of private equity investment on entry in high-entryindustries is robust to accounting for other industry characteristics, as well as for othercharacteristics of the business environment that have been suggested by the literature asdeterminants of entry rates.ECBWorking Paper Series No 1078August 200931

This paper is, to our knowledge, the …rst attempt to empirically link private equityinvestment to industry entry in a cross-country cross-industry setting. Our results imply thatprivate equity in general and venture capital in particular have a positive e¤ect in potentiallybringing new ideas to the marketplace in the shape of young companies. In this paper we arenot interested in the pro…tability of this enterprise, but as far as real e¤ects are concerned,private equity investment seems to generate value through fostering entrepreneurial activityin the economy. A number of important questions remain unanswered due to the natureof our data.For example, what is the relative importance of the di¤erent channels viawhich private equity and venture capital a¤ect entry? Is it more anticipatory considerationswhere nascent entrepreneurs are aware of future …nancing needs and are more likely todecide to start their own business at a time when more venture capital is available? Or isit entrepreneurial spawning where venture-backed …rms are more likely to create other newventures? Future research can greatly contribute by addressing those questions.32 ECBWorking Paper Series No 1078August 2009

References[1] Aernoudt, R., 1999. European policy towards venture capital: myth or reality? VentureCapital, 1(1), 47-58.[2] Aghion, P., T. Fally, and S. Scarpetta, 2007. Credit constraints as a barrier to the entryand post-entry growth of …rms. Economic Policy, 22, 731-779.[3] Aghion, P., and P. Howitt, 1992. A model of growth through creative destruction.Econometrica, 60(2), 323-351.[4] Aghion, P., and P. Howitt, 1998. Endogenous Growth Theory. MIT Press: Cambridge,MA.[5] Barro, R., and J.-W. Lee, 2000. International data on educational attainment: updatesand implications. CID Working Paper No. 42.[6] Beck, T., A. Demirgüç-Kunt, and V. Maksimovic, 2005. Financial and legal constraintsto growth: does …rm size matter? Journal of Finance, 60(1), 137-177.[7] Beck, T., A. Demirguc-Kunt, L. Laeven, and R. Levine, 2008. Finance, …rm size, andgrowth. Journal of Money, Credit, and Banking, 40(7), 1379-1405.[8] Black, S.E., and P.E. Strahan, 2002. Entrepreneurship and bank credit availability.Journal of Finance, 57(6), 2807-2833.[9] Bresnahan, T., and P. Reiss, 1988. Do entry conditions vary across markets? BrookingsPapers on Economic Activity 3, 833-881.[10] Botazzi, L., M. Da Rin, and T. Hellmann, 2008. Who are the active investors? Evidencefrom venture capital. Journal of Financial Economics, 89(3), 488-512.[11] Caves, R., M. Winston, and M. Hurwitz, 1991. Patent expiration, entry, and competitionin the US pharmaceutical industry. Brookings Papers on Economic Activity 6, 1-66.ECBWorking Paper Series No 1078August 200933

[12] Cetorelli, N., and P.E. Strahan, 2006. Finance as a barrier to entry: bank competitionand industry structure in local U.S. markets. Journal of Finance, 61(1), 437-461.[13] Christensen, Clayton M., 1997. The Innovator’s Dilemma: When New TechnologiesCause Great Firms to Fail. Harvard Business School Press: Boston, MA.[14] Claessens, S., and L. Laeven, 2003. Financial development, property rights, and growth.Journal of Finance, 58(6), 2401-2436.[15] Da Rin, M., Di Giacomo, M., and A. Sembenelli, 2008. Firm entry dynamics and thetaxation of corporate pro…ts: evidence from Europe. CentER Discussion Paper SeriesNo. 2008-65.[16] Davis, S.J., J. Haltiwanger, J.Lerner, and J. Miranda, 2008. Private equity and employment.U.S. Census Bureau Center for Economic Studies Paper No. CES-WP-08-07.[17] De Soto, H., 1990. The Other Path. Harper and Row: New York, NY.[18] Enste, D., and F. Schneider, 2000. Shadow economies: Size, causes, and consequences.Journal of Economic Literature, 38(1), 77-114.[19] Geroski, P.A., 1995. What do we know about entry? International Journal of IndustrialOrganization, 13(4), 421-440.[20] Gilson, R. J., 2003. Engineering a venture capital market: lessons from the Americanexperience. Stanford Law Review, 55(4), 1067-1103.[21] Gompers, P.A., and J. Lerner, 1999. The Venture Capital Cycle. MIT Press: Cambridge,MA.[22] Gompers, P.A., and J. Lerner, 2001a. The Money of Invention. Harvard Business SchoolPress: Boston, MA.[23] Gompers, P.A., and J. Lerner, 2001b. The venture capital revolution. Journal of EconomicPerspectives, 15 (2), 145-168.34 ECBWorking Paper Series No 1078August 2009

[24] Gompers, P.A., J. Lerner, and D.S. Scharfstein, 2005. Entrepreneurial spawning: publiccorporations and the formation of new ventures, 1986-1999. Journal of Finance, 60(2),577-614.[25] Hellmann, T., 1998. The allocation of control rights in venture capital contracts. RANDJournal of Economics, 29(1), 57-76.[26] Hellmann, T., and M. Puri, 2000. The interaction between product market and …nancingstrategy: the role of venture capital. Review of Financial Studies, 13(4), 959-984.[27] Klapper, L., L. Laeven, and R. Rajan, 2006. Entry regulation as a barrier to entrepreneurship.Journal of Financial Economics, 82(3), 591-629.[28] Kaplan, S., and P. Stromberg, 2001. Venture capitalists as principals: contracting,screening, and monitoring. American Economic Review, 91(2), 426-430.[29] Kortum, S., and J. Lerner, 2000. Assessing the contribution of venture capital to innovation.RAND Journal of Economics, 31(4), 674-692.[30] La Porta, R., F. Lopez de Silanes, A. Shleifer, and R. Vishny, 1998. Finance and law.Journal of Political Economy, 106(6), 1113-55.[31] Lafontaine, F., and J. Sivadasan, 2008. Do labor market rigidities have microeconomice¤ects?Evidence from within the …rm. American Economic Journal: Applied Economics,Forthcoming.[32] Mollica, M., and L. Zingales, 2007. The impact of venture capital on innovation and thecreation of new businesses. University of Chicago Working paper.[33] Park, W., and J.C. Ginarte, 1997. Intellectual property rights and economic growth.Contemporary Economic Policy, 15(3), 51-61.[34] Perotti, E., and P. Volpin, 2007. Investor protection and entry. Tinbergen InstituteDiscussion Paper 07-006/2.ECBWorking Paper Series No 1078August 200935

[35] Petersen, M., and R. Rajan, 1995. The e¤ect of credit market competition on lendingrelationships. The Quarterly Journal of Economics, 110(2), 407-443.[36] Rajan, R., and L. Zingales, 1998. Financial dependance and growth. American EconomicReview, 88(3), 559-586.[37] Rajan, R., and L. Zingales, 2003. The great reversal: the politics of …nancial developmentin the 20th century. Journal of Financial Economics, 69(1), 5-50.[38] Schoar, A., 2002. E¤ects of corporate diversi…cation on productivity. Journal of Finance,57(6), 2379-2403.[39] Schumpeter, Joseph A. The Process of Creative Destruction. Unwin. 1942.[40] Stein, Jeremy C., 2002. Information production and capital allocation: decentralizedversus hierarchical …rms. Journal of Finance, 57(5), 1891-1921.[41] Tang, M.-C., and Y.-L. Chyi, 2008. Legal environments, venture capital, and total factorproductivity. Contemporary Economic Policy, 26, 468-481.36 ECBWorking Paper Series No 1078August 2009

Table 1Summary statistics: EVCA data on private equity investmentCountryTotal PE investmentover GDP, in %Total PE investmentover GDP, in %Austria 0.027 0.079Belgium 0.119 0.084Czech Republic 0.042 0.023Denmark 0.029 0.353Finland 0.111 0.149France 0.117 0.375Germany 0.081 0.145Greece 0.027 0.004Hungary 0.052 0.086Italy 0.082 0.131Netherlands 0.271 0.403Norway 0.112 0.164Poland 0.093 0.064Spain 0.062 0.265Sweden 0.200 0.820UK 0.465 1.215The data are averaged for the period 1998-99 (columns 2) and for 2006-2007 (column 3), and areaggregated over industries. The respective values include all private equity investment in the respectiveEuropean country by European banks and private equity houses, as well as from international banksand private equity houses which have a European branch or partner. Source: EVCA yearbooks.ECBWorking Paper Series No 1078August 200937

Table 2Summary statistics on entry, by countryNew firms (

Table 3Entry rates by industry in Europe against the US benchmarkNew firms as %of all firms, 1998-1999, EuropeNew firms as% of all firms,1998-1999, US2-digit NACE rev 1.1 code and industry nameManufacturing15. Food products and beverages 9.6 10.416. Tobacco products 16.1 14.817. Textiles 9.3 13.718. Wearing apparel; dressing and dying of fur 9.5 12.819. Tanning and dressing of leather; luggage, handbags, saddlery,harness, and footwear 8.6 18.220. Wood and products of wood and cork, except furniture 11.2 1221. Pulp, paper and paper products 9.5 10.622. Publishing, printing and reproduction of recorded media 11.2 1123. Coke, refined petroleum products, and nuclear fuel 10.8 11.524. Chemicals and chemical products 9.3 12.225. Rubber and plastic products 10.9 926. Other non-metallic mineral products 9.3 11.627. Basic metals 12.2 9.828. Fabricated metal products, except machinery and equipment 11.5 11.429. Machinery and equipment not elsewhere classified 10.4 8.630. Office machinery and computers 15.8 17.431. Electrical machinery and apparatus not elsewhere classified 11.0 11.832. Radio, television, and communication equipment and apparatus 13.8 16.533. Medical, precision, and optical instruments, watches, and clocks 9.7 11.434. Motor vehicles, trailers, and semi-trailers 10.8 10.435. Other transport equipment 13.0 1636. Furniture; manufacturing not elsewhere classified 11.7 15.8Construction45. Construction 13.7 16.3Trade50. Sale, maintenance and repair of motor vehicles andmotorcycles; retail sale of automotive fuel 13.4 10.251. Wholesale trade and commission trade, except of motor vehicles 14.9 10.652. Retail trade, except of motor vehicles and motorcycles; repair ofpersonal and household goods 15.2 14.4Hotels and restaurants55. Hotels and restaurants 14.8 11.8Transportation60. Land transport; transport via pipelines 16.0 16.861. Water transport 11.9 11.262. Air transport 13.5 12.463. Supporting and auxiliary transport activities, and travel agencies 14.5 13.5Services70. Real estate activities 16.0 10.671. Renting of machinery and equipment without operator and ofpersonal and household goods 18.0 12.672. Computer and related services 22.4 21.473. Research and development 17.0 13.174. Other business activities 18.4 19.4Other93. Other services activities 17.0 13The values reported are averaged over 1998 and 1999, for Europe (column 2) and the US (column 3). Thefraction of new firms is calculated as the number of firms 2 or less years old over all firms in this particularindustry. Source: Amadeus database (column 2) and Dun and Bradstreet (column 3)ECBWorking Paper Series No 1078August 200939

Table 42-digit NACE rev. 1.1 industry classification -> EVCA industry classification conversion keyNACE rev 1.1 industry classEVCA industry class22, 32 Communications24, 30, 32, 33, 72 Computer73 Biotechnology23, 33, 85 Medical and health related23, 24 Energy15, 16, 17, 18, 19, 34, 35, 36, 50, 51, 52, 55, 63, 93 Consumer products26, 27, 28, 29 Industrial products24, 25 Chemicals and chemical related31 Industrial automation20, 21 Other manufacturing60, 61, 62 Transportation61, 63, 71, 74 Other services45 Construction70 OtherSource: EVCA -> NACE rev. 2 and NACE rev. 2 -> NACE rev. 1.1 translation key.40 ECBWorking Paper Series No 1078August 2009

Table 5Private equity and firm entry: measuring PE investment at the country levelIndustry shareEntry US * Private equity(i)OLS(ii)Tobit(iii)TobitFraction of new firms Fraction of exited firmsPE investment PE investment Excluding trans. economies Small firms only PE investment-0.005 -0.005 0.156 -0.026 -0.012(0.09) (0.109) (0.116) (0.055) (0.101)0.300 0.300 0.189 0.263 0.239(0.112)** (0.078)*** (0.079)** (0.117)** (0.059)***Observations 571 571 471 571 571R 2 0.60 0.56 0.63(iv)OLS(v)OLSThe reported estimates are from OLS regressions (columns (i) and (iv)-(v)) and from a Tobit regression (columns (ii) and (iii)). The dependent variable is theratio of all new firms to total firms (in columns (i)-(iii)), the ratio of all new firms with less than 10 employees to total firms (column (iv)), and the ratio ofall exited firms to total firms by 2-digit NACE rev. 1.1, averaged for 1998-1999 (column (v)). Industry share is the ratio of industry sales to total sales by allindustries in each country. Entry US is the ratio of new firms to total firms in the US, by 2-digit NACE rev. 1.1. Private equity measures actual private equityinvestment at the country level, normalized by GDP, averaged for 1998-1999. All regressions include a constant, 2-digit industry dummies and countrydummies, not reported. White’s heteroskedasticity corrected standard errors standard errors are reported in parentheses. ***, **, and * report significance atthe 1%, 5% and 10% level, respectively. See Appendix 1 for detailed variable definitions and sources.ECBWorking Paper Series No 1078August 200941

Table 6Private equity timing and firm entryIndustry shareEntry US * Private equity(i)OLSPE investment, 3-yearaverage(ii)OLS(iii)OLSAll firms Small firmsPE investment, 4-yearaveragePE investment, 3-yearaverage(iv)OLSPE investment, 4-yearaverage-0.004 -0.003 -0.024 -0.023(0.091) (0.091) (0.054) (0.054)0.349 0.390 0.313 0.362(0.131)*** (0.156)*** (0.138)** (0.163)**Observations 571 571 571 571R 2 0.60 0.60 0.56 0.56The reported estimates are from OLS regressions. The dependent variable is the ratio of all new firms to total firms (columns (i)-(ii)) and the ratio of all newfirms with less than 10 employees to total firms (columns (iii)-(iv)), by 2-digit NACE rev. 1.1 and averaged for 1998-1999. Industry share is the ratio ofindustry sales to total sales by all industries in each country. Entry US is the ratio of new firms to total firms in the US, by 2-digit NACE rev. 1.1. Privateequity measures actual private equity investment at the country level, normalized by GDP, and averaged over 1997-1999 (columns (i) and (iii)) and 1996-1999 (columns (ii) and (iv)). All regressions include a constant, 2-digit industry dummies and country dummies, not reported. White’s heteroskedasticitycorrected standard errors standard errors are reported in parentheses. ***, **, and * report significance at the 1%, 5% and 10% level, respectively. SeeAppendix 1 for detailed variable definitions and sources.42 ECBWorking Paper Series No 1078August 2009

Table 7Alternative proxies for propensity to enterIndustry share(i) (ii) (iii) (iv) (iv) (vi)Fraction of new firmsFirm exit rates, US SME entry rates, US Firm entry rates, UKAll firms Small firms All firms Small firms All firms Small firms-0.002 -0.018 -0.036 -0.003 0.004 0.022(0.119) (0.072) (0.118) -0.118 (0.119) -0.1180.018 0.029Exit US * Private equity(0.026) (0.017)*Entry by SMEs US *0.234 0.207Private equity (0.107)** (0.107)**Entry UK * Private equity0.271 0.086(0.106)*** (0.094)Observations 571 571 571 571 534 534R 2 0.60 0.61 0.60 0.55 0.60 0.67The reported estimates are from OLS regressions. The dependent variable is the ratio of new firms to total firms (columns (i), (iii), and (v)) and of newfirms with less than 10 employees to total firms (columns (ii), (iv), and (vi)), by 2-digit NACE rev. 1.1 and averaged for 1998-1999. Industry share is theratio of industry sales to total sales by all industries in each country. Exit US is the ratio of firms that ceased to exist to total firms in the US. Entry bySMEs US measures entry rates by small or medium enterprises only (

Table 8Private equity and alternative proxies for financeIndustry share(i) (ii) (iii) (iv) (v) (vi) (vii) (viii)All firms Small firmsPrivate creditover GDPAccess tofinanceInvestors’protectionHorse race Private creditover GDPAccess tofinanceInvestors’protectionHorse race-0.015 -0.010 -0.003 -0.008 -0.022 -0.03 -0.022 -0.019(0.109) (0.108) (0.109) (0.108) (0.069) (0.068) (0.067) (0.068)Entry US * Private equity 0.232 0.207 0.237 0.217 0.317 0.295 0.131 0.169(0.093)*** (0.085)*** (0.100)** (0.128)* (0.059)*** (0.054)*** (0.065)** (0.082)**Entry US *0.583 0.195 0.375 0.072Private credit over GDP (0.333)* (0.409) (0.222)* (0.267)Entry US *0.028 0.029 0.014 0.545Access to finance (0.011)*** (0.012)** (0.006)* (0.795)Entry US *0.198 0.209 0.242 0.221Investors’ protection (0.118)* (0.126)* (0.078)*** (0.083)***Observations 571 571 571 571 571 571 571 571R 2 0.63 0.69 0.68 0.69 0.64 0.64 0.64 0.64The reported estimates are from OLS regressions. The dependent variable is the ratio of new firms to total firms (columns (i)-(iv)) and of new firms with lessthan 10 employees to total firms (columns (v)-(viii)), by 2-digit NACE rev. 1.1 and averaged for 1998-1999. Industry share is the ratio of industry sales tototal sales by all industries in each country. Entry US is the ratio of new firms to total firms in the US, by 2-digit NACE rev. 1.1. Private credit over GDPmeasures the ratio of private credit allocated by commercial banks to country total GDP. Access to finance measures general access to external finance byhouseholds and businesses. Investor’s protection measures the degree of legal protection of private investment in the country. All regressions include aconstant, 2-digit industry dummies and country dummies, not reported. White’s heteroskedasticity corrected standard errors standard errors are reported inparentheses. ***, **, and * report significance at the 1%, 5% and 10% level, respectively. See Appendix 1 for detailed variable definitions and sources.44 ECBWorking Paper Series No 1078August 2009

Table 9Endogeneity and selection issuesIndustry shareEntry US * Private equityHigh legal efficiency *(i) (ii) (iii) (iv) (v) (vi) (vii) (viii) (ix) (x)All firms Small firms onlyPE lawsand fundsizeSmallindustriesonlyHigh-lowlegalefficiencyHigh-lowshadoweconomy PE lawsPE lawsand fundsizeSmallindustriesonlyHigh-lowlegalefficiencyHigh-lowshadoweconomyPE laws0.001 -0.019 0.328 -0.095 -0.017 -0.016 0.019 0.082 -0.031 -0.034(0.091) (0.097) (0.239) (0.092) (0.091) (0.053) (0.059) (0.157) (0.059) (0.065)0.253 0.242 0.276 0.234 0.293 0.264(0.144)* (0.121)** (0.120)** (0.143)* (0.135)** (0.126)**0.268 0.275Entry US * Private equity (0.121)** (0.122)**Low legal efficiency *-0.435 0.179Entry US * Private equity (0.472) (0.302)Low share shadow econ *0.278 0.262Entry US * Private equity (0.122)** (0.122)***High share shadow econ *0.283 0.174Entry US * Private equity (0.176) (0.127)Observations 571 571 515 438 443 571 571 515 438 443R 2 0.60 0.60 0.59 0.66 0.67 0.56 0.57 0.54 0.63 0.68The reported estimates are from an IV regression (columns (i)-(ii) and (vi)-(vii)) and from OLS regressions (columns (iii)-(v) and (viii)-(x)). The dependentvariable is the ratio of new firms to total firms (columns (i)-(v)) and the ratio of new firms with less than 10 employees to total firms (columns (vi)-(x)), by 2-digit NACE rev. 1.1 and averaged for 1998-1999. Private equity measures actual private equity investment at the country level, normalized by GDP. Privateequity laws is a dummy equal to 1 if by 1998 pension funds were allower to invest in risk capital markets in the respective country. In columns (ii) and (vi), thedummy is interacted with the actual size of pension funds assets as share of GDP. Columns (iii) and (viii) perform the analysis on the sample of all industriesexcluding the top 10% in terms of share of sales. Industry share is the ratio of industry sales to total sales by all industries in each country. Entry US is the ratioof new firms to total firms in the US, by 2-digit NACE rev. 1.1. High(low) legal efficiencies are dummies equal to 1 if the country is in the top(bottom) 33% interms of legal efficiency, and to 0 otherwise. High(low) share shadow economy are dummies equal to 1 if the country is in the top(bottom) 33% in terms ofimplied size of the unofficial economy, and to 0 otherwise. All regressions include a constant, 2-digit industry dummies and country dummies, not reported.White’s heteroskedasticity corrected standard errors standard errors are reported in parentheses. ***, **, and * report significance at the 1%, 5% and 10%level, respectively. See Appendix 1 for detailed variable definitions and sources.ECBWorking Paper Series No 1078August 200945

Appendix 1. Variables: definitions and sourcesVariableDefinition and sourceEVCA dataPrivate equityActual PE investment by European banks, mutual funds and PE houses, as well asby international PE investors with a European branch or partner, normalized byGDP; by country or European Venture Capital Association (EVCA) industryclassification. Source: EVCA yearbooksSeed and start-up finance Actual PE investment by European banks, mutual funds and PE houses, as well asby international PE investors with a European branch or partner, normalized byGDP, allocated to seed and start-up firms; by country. Source: EVCA yearbooksSeed,start-up & expansion Actual PE investment by European banks, mutual funds and PE houses, as well asby international PE investors with a European branch or partner, normalized byGDP, allocated to all private equity deals expect for buyouts; by country. Source:EVCA yearbooksIndependent PENon-independent PEPrivate equity investment by independent PE funds, normalized by GDP, allocatedto all private equity deals; by country. Source: EVCA yearbooksPrivate equity investment by captive, semi-captive and public PE funds,normalized by GDP, allocated to all private equity deals; by country. Source:EVCA yearbooksAmadeus dataEntryNumber of firms 2 years or younger as a fraction of the total firms. Averaged for1998-1999 and 2006-2007. Used at the 2-digit NACE rev. 1.1 level andaggregated weighted by employment at the EVCA industry classification level.Source: AmadeusExit Number of exited firms as a fraction of the total firms. Averaged for 1998-1999and 2006-2007. Used at the 2-digit NACE rev. 1.1 level and aggregated weightedby employment at the EVCA industry classification level. Source: AmadeusEntry by small firmsNumber of firms 2 years or younger with less than 10 employees as a fraction ofthe total firms. Averaged for 1998-1999 and 2006-2007. Used at the 2-digit NACErev. 1.1 level and aggregated weighted by employment at the EVCA industryclassification level. Source: AmadeusIndustry share Fraction of the industry’s sales in total sales. Averaged for 1998-1999 and 2006-2007. Used at the 2-digit NACE rev. 1.1 level and aggregated weighted byemployment at the EVCA industry classification level. Source: AmadeusUS benchmarkEntry USEntry rates for US corporations. Calculated for 2-digit NACE industries (originaldata on a 4-digit SIC level). Average for the years 1998-99. Source: Dun &Bradstreet.46 ECBWorking Paper Series No 1078August 2009

Exit USExit rates for US corporations. Calculated for 2-digit NACE industries (originaldata on a 4-digit SIC level). Average for the years 1998-99. Source: Dun &Bradstreet.Entry by SMEs US Entry rates for US corporations with less than 250 employees. Calculated for 2-digit NACE industries (original data on a 4-digit SIC level). Average for the years1998-99. Source: Dun & Bradstreet.Entry UKCapital intensityR&D intensityFinancial dependenceIndustry growthEntry rates for UK corporations. Calculated for 2-digit NACE industries (originaldata on a 4-digit SIC level). Average for the years 1998-99. Source: Dun &Bradstreet.Measure of physical capital usage, equal to the industry-level median of the ratioof physical capital used to sales. The numerator and denominator are summed overall years for each firm before dividing. Computed for all U.S. firms for the period1990-99. Calculated for 2-digit NACE industries (original data on a 4-digit SIClevel). Source: Compustat.Measure of dependence on research and development, equal to the industry-levelmedian of the ratio of R&D expenses to sales. The numerator and denominator aresummed over all years for each firm before dividing. Computed for all U.S. firmsfor the period 1990-99. Calculated for 2-digit NACE industries (original data on a4-digit SIC level). Source: Compustat.Industry-level median of the ratio of capital expenditures minus cash flow overcapital expenditures. The numerator and denominator are summed over all yearsfor each firm before dividing. Cash flow is defined as the sum of funds fromoperations, decreases in inventories, decreases in receivables, and increases inpayables. Capital expenditures include net acquisitions of fixed assets. Thisdefinition follows Rajan and Zingales (1998). We compute this measure for allU.S. firms for the period 1990-99. Calculated for 2-digit NACE industries(original data on a 4-digit SIC level). Source: Compustat.Measure of industry growth, equal to the industry-level median of value added perworker. The numerator and denominator are summed over all years for each firmbefore dividing. Computed for all U.S. firms for the period 1990-99. Calculatedfor 2-digit NACE industries (original data on a 4-digit SIC level). Source:Compustat.Country-level variablesPrivate credit over GDPAccess to financeInvestors’ protectionRatio of domestic credit to the private sector scaled by GDP, average over theperiod 1995-99. Source: International Monetary Fund’s International FinancialStatistics (IMF-IFS).General index of access to external finance by households and businesses. Source:Finance for All? The World Bank, 2007.Index of the degree of protection of investors, calculated as an average of threeindices: transparency of transactions, liability for self-dealing, and shareholders’ability to sue officers and directors for misconduct; averaged over 1995-1999.Source: Doing Business Database (WB).ECBWorking Paper Series No 1078August 200947

Share shadow economyShare of the informal economy, calculated as the size of the informal economy asa percentage of official GNI. Average over the period 1999-2000. Source: Ensteand Schneider (2000).Property rights protection Index of degree of protection of intellectual property rights. Source: Park andGinarte (1997).Entry proceduresNumber of procedures to register a business. Data for the year 1999. Source:Djankov et al. (2002).Labor regulations Index of the legal ease of hiring and firing workers; averaged over 1995-1999.Source: Doing Business Database.Profit taxHuman capitalOutput gapPension funds sizeMeasure of the marginal tax on profit. Source: PriceWaterHouseCoopersWorldwide Taxes 1999-2000).Average years of schooling for an individual in the respective country. Source:Barro and Lee “International Data on Educational Attainment” dataset.Difference between actual and potential GDP, averaged for 1998-1999 and2005/2006. Source: Econ stats online, http://www.econstats.com/weo/V027.htmAssets held by all pension funds in the respective country normalized by GDP andaveraged for 1998-1999. Source: Eurostat48 ECBWorking Paper Series No 1078August 2009

European Central Bank Working Paper SeriesFor a complete list of Working Papers published by the ECB, please visit the ECB’s website(http://www.ecb.europa.eu).1049 “Labour force participation in the euro area: a cohort based analysis” by A. Balleer, R. Gómez-Salvadorand J. Turunen, May 2009.1050 “Wealth effects on consumption: evidence from the euro area” by R. M. Sousa, May 2009.1051 “Are more data always better for factor analysis? Results for the euro area, the six largest euro area countriesand the UK” by G. Caggiano, G. Kapetanios and V. Labhard, May 2009.1052 “Bidding behaviour in the ECB’s main refinancing operations during the financial crisis” by J. Eisenschmidt,A. Hirsch and T. Linzert, May 2009.1053 “Inflation dynamics with labour market matching: assessing alternative specifications” by K. Christoffel, J. Costain,G. de Walque, K. Kuester, T. Linzert, S. Millard and O. Pierrard, May 2009.1054 “Fiscal behaviour in the European Union: rules, fiscal decentralization and government indebtedness”by A. Afonso and S. Hauptmeier, May 2009.1055 “The impact of extreme weather events on budget balances and implications for fiscal policy” by E. M. Lisand C. Nickel, May 2009.1056 “The pricing of subprime mortgage risk in good times and bad: evidence from the ABX.HE indices” by I. Fenderand M. Scheicher, May 2009.1057 “Euro area private consumption: Is there a role for housing wealth effects?” by F. Skudelny, May 2009.1058 “National prices and wage setting in a currency union” by M. Sánchez, May 2009.1059 “Forecasting the world economy in the short-term” by A. Jakaitiene and S. Dées, June 2009.1060 “What explains global exchange rate movements during the financial crisis?” by M. Fratzscher, June 2009.1061 “The distribution of households consumption-expenditure budget shares” by M. Barigozzi, L. Alessi, M. Capassoand G. Fagiolo, June 2009.1062 “External shocks and international inflation linkages: a global VAR analysis” by A. Galesi and M. J. Lombardi,June 2009.1063 “Does private equity investment spur innovation? Evidence from Europe” by A. Popov and P. Roosenboom,June 2009.1064 “Does it pay to have the euro? Italy’s politics and financial markets under the lira and the euro” by M. Fratzscherand L. Stracca, June 2009.1065 “Monetary policy and inflationary shocks under imperfect credibility” by M. Darracq Pariès and S. Moyen,June 2009.1066 “Universal banks and corporate control: evidence from the global syndicated loan market” by M. A. Ferreiraand P. Matos, July 2009.1067 “The dynamic effects of shocks to wages and prices in the United States and the euro area” by R. Duarteand C. R. Marques, July 2009.ECBWorking Paper Series No 1078August 200949

1068 “Asset price misalignments and the role of money and credit” by D. Gerdesmeier, H.-E. Reimers and B. Roffia,July 2009.1069 “Housing finance and monetary policy” by A. Calza, T. Monacelli and L. Stracca, July 2009.1070 “Monetary policy committees: meetings and outcomes” by J. M. Berk and B. K. Bierut, July 2009.1071 “Booms and busts in housing markets: determinants and implications” by L. Agnello and L. Schuknecht, July 2009.1072 “How important are common factors in driving non-fuel commodity prices? A dynamic factor analysis”by I. Vansteenkiste, July 2009.1073 “Can non-linear real shocks explain the persistence of PPP exchange rate disequilibria?” by T. Peltonen, M. Sagerand A. Popescu, July 2009.1074 “Wages are flexible, aren’t they? Evidence from monthly micro wage data” by P. Lünnemann and L. Wintr,July 2009.1075 “Bank risk and monetary policy” by Y. Altunbas, L. Gambacorta and D. Marqués-Ibáñez, July 2009.1076 “Optimal monetary policy in a new Keynesian model with habits in consumption” by C. Leith, I. Moldovanand R. Rossi, July 2009.1077 “The reception of public signals in financial markets – what if central bank communication becomes stale?”by M. Ehrmann and D. Sondermann, August 2009.1078 “On the real effects of private equity investment: evidence from new business creation” by A. Popovand P. Roosenboom, August 2009.50 ECBWorking Paper Series No 1078August 2009